Error-rate Estimation with Ones Counting

Size: px
Start display at page:

Download "Error-rate Estimation with Ones Counting"

Transcription

1 UC TCHNICA RPORT CNG-8- ng Hseh epatment of lectcal ngneeng teb chool of ngneeng Unvesty of ouen Calfona o-ate stmaton w Ones Countng Zhaolang Pan elvn A Beue Abstact W I ccut featue sze scalng down, t s becomng moe dffcult expensve to acheve a desed level of yeld o-toleance, at s, employng defectve chps at occasonally poduce eoneous yet acceptable esults n tageted applcatons, has been poposed as one way to ncease effectve yeld In e doman of eo-toleance, defectve chps ae chaactezed by cetan ctea set by vaous applcatons o-ate, namely how fequent eos occu at e output, s one such cteon In s epot we focus on e followng poblem: Gven a combnatonal logc ccut at s defectve hence occasonally poduces an eoneous output, how can one detemne e eo-ate of each output lne by usng ones countng? Ths poblem was pevously consdeed by oes but e analyss has some naccuaces Ths epot pesents bo a moe accuate analyss new estmato of eo-ate C I INTROUCTION lasscal I post-manufactung tests patton chps nto two categoes, namely ose at pass ose at fal Pesumably, chps at fal have defects ae dscaded, whle e oes ae sold to customes In s epot we wll use e tems fal defectve chp ntechangeably The facton of defect-fee chps among all manufactued chps s called e yeld of a I manufactung pocess W mnmal featue szes scalng down to below 9nm, t s becomng moe dffcult, hence expensve, to acheve a desed level of yeld [] efectve chps contbute to a lage economc loss to chp manufactues, nceased pces to consumes In an attempt to ncease yeld, desgnes employ seveal technques such as fault-toleance, defect-toleance desgn-fo-manufactuablty [] [] [] Anoe somewhat adcal new technque oogonal to ese s called eotoleance [5] [6] A chp s sad to be eo-toleant T w espect to an applcaton o system f t contans defects hence occasonally poduces eos, usng s chp n a gven applcaton/system poduces acceptable esults to e Ths epot s based upon wok suppoted n pat by e Natonal cence Foundaton unde Gant No 89 end use Because a facton of defectve chps ae used n systems ae an beng dscaded, eo-toleance nceases effectve yeld, hence potentally poft any examples of systems at can opeate n an eotoleant mode esde n e doman of mult-meda manmachne ntefaces ome smple applcatons ae befly descbed below Whle e applcaton of eo-toleance n dgtal systems pobably dates back seveal decades, one of e ealest documented applcatons we have encounteed appeas n a 995 patent: It s heen ecognzed at n an audo sgnal pocessng system, a RA chp fo stong dgtzed audo sgnals selected to nclude at least one nopeatve memoy locaton, s acceptable fo use as a stoage medum n at no notceable eo s poduced on playback of e ecoded sgnal due to e samplng ate of e audo sgnal due to e elatvely low ate of defects allowed Fuemoe, e use of a lessan-pefect RA chp fo stong audo nfomaton s acceptable due to e substantally non-ctcal natue of audo sgnals, as opposed to e extemely ctcal natue of compute data [7] The followng statement appeas n anoe patent dealng w a telephone-answeng machne: The use of audo RAs ARAs s also known n e answeng machne at, w ARAs beng RAs at ae allowed to have a small numbe of defectve bts, n ode to allow e use of lowecost ntegated ccut memoy chps [8] Bo statements embody e concept of eo toleance, e, at some defectve ccuty can poduce acceptable pefomance An PG ccut s anoe example whee eo-toleance s applcable The effect of defects n some functonal blocks of a PG encode, such as e moton estmaton block, on e pefomance of e ccuts has been studed [9] [] The esults ndcate at fo about 5% of ose defects at can be modeled as a sngle o double stuck-at fault, e esultng defectve PG ccuts geneate acceptable pefomance Oe examples of eo-toleant applcatons have been epoted elsewhee [] [5] o-toleant chps ange fom low pced memoes to hgh pced audo vdeo pocessng chps In e nea futue, many system-on-chp

2 UC TCHNICA RPORT CNG-8- poducts wll be excellent cdates fo eo-toleant applcatons, such as ose contanng embedded language tanslatos as well as audo, touch, smell /o vdeo pocessos Gven an analog system, one can nject eos nto e ccut at some node by applyng nose at at node measue attbutes of e output, such as db loss, O value, o nstablty In a dgtal ccut, one can ntoduce eos on an ntenal sgnal lne measue e esultng mpact on e data at e output bus Two measues at have been used to quantfy output eos ae eo-ate eo-sgnfcance [5] [6] A fomal defnton of eo-ate s povded n [] o-ate ndcates how often on aveage an eo s seen at e output when om nput pattens ae appled If e ccut s combnatonal contans a defect, e pattens ae unfomly selected ove e nput space, en e eo-ate s equvalent to e facton of e nput space at detects e defect, e, poduces an eoneous output o-sgnfcance ndcates e magntude of an eo, when e assocated patten epesents some numec quantty ependng on e applcaton ee one o bo of ese measues mght be applcable R C R Fg mplfed BIT achtectue In s epot we focus on estmatng e eo-ate of a defectve ccut havng statc defects Based on e eoate, defectve chps can be bnned nto categoes Fo example, two eo-ate esholds, say, whee < < <, can be used to dvde e ange of eo-ate nto ee pats, e,, ], < ], ] efectve chps can now be pattoned nto ee categoes, namely categoy I w eo-ate n e ange, ], categoy II n < ], categoy III n, ] Chps n categoy III hgh eo-ate ae dscaded; chps n categoy II ae sold at a lage dscount n pce; ose n categoy I ae sold at a modeate dscount Ou technque fo estmatng eo-ate eques at we apply a lage numbe of pseudo-om pattens to a tageted ccut ost complex moden ccuts employ ee a fullscan FT o a BIT meodology [6] [7] Whle ee of ese test meodologes can be used to mplement ou eoate estmaton technque, we wll descbe ou wok n e context of a BIT meodology snce s smplfes e dscusson A smplfed veson of a BIT achtectue used O efes to mean opnon scoe, ndcates e qualty of audo speech afte pocessng by some COC to test combnatonal logc s shown n Fg In nomal opeaton, egste R eceves data n paallel dves e block of logc C C n tun passes data to R In e desgnfo-test pocess, ese egstes ae modfed to opeate dffeently when n e test mode In e test mode, R opeates as a pseudo-om patten geneato PRPG, R as a multple nput sgnatue analyze IR The opeaton of e test pocess s as follows Fst e egstes ae ntalzed to some known states Then e egstes ae clocked tmes, whee s often a vey lage numbe At e end of s pocess, f e state of e IR s e same as some pe-computed value, en e ccut C s sad to pass s good o defect fee, oewse t fals The dea hee s at f ee s a defect n e ccut, fo at least one of e nput pattens e coespondng output patten would be n eo, once an eo entes e IR, w a vey hgh pobablty e state of e IR emans n eo even as new data aves at ts nputs W some assumptons egadng e dstbuton of output pattens, s technque can be effectvely appled to some foms of sequental ccuts, such as feed-fowad ppelnes Thus, to mplement e bnnng pocess, t s mpotant to effcently bound e eo-ate of a defectve chp Note at e test meodology dscussed n s epot does not employ a fault model, test patten geneaton o fault smulaton A technque fo estmatng e eo-ate of a block of logc usng sgnatue analyss was fst poposed n [] ubsequently, Pan Beue sgnfcantly exped on e eoetcal aspects of s meodology [] hahd Gupta en poposed a vaant of s wok usng ones countng ae an sgnatue analyss [] Unfotunately, e analyss mpled at eo-ate estmaton va ones countng used sgnfcantly moe esouces an at equed va sgnatue analyss As wll be explaned late, e esults ae counte ntutve In s epot, we befly evew e analyss pesented n [], pesent a new analyss fo s ones countng technque a new estmato of eo-ate The esults fom ou analyss show at ones countng sgnatue analyss technques use compaable esouces fo obtanng e same degee of accuacy n eo-ate estmaton Ths epot s oganzed as follows ecton evews e ones countng eo-ate estmaton wok pesented n [] In ecton, we pesent ou new analyss of s technque A new estmato s poposed e statstcal chaactestcs of e estmato ae pesented eveal specal cases of usng e ones countng technque ae descbed n ecton In ecton 5, smulaton esults ae pesented The poblem of chp classfcaton based on ones countng eo-ate estmaton s dscussed n ecton 6 ecton 7 pesents e expemental esults of chp classfcaton, ecton 8 concludes e epot II RIW OF PRIOU WORK The eo-ate estmaton technques poposed n [5], [6],

3 UC TCHNICA RPORT CNG-8- [] [] ae llustated usng a BIT achtectue, ough a scan achtectue s equally applcable The egste dvng e ccut-unde-test CUT s a PRPG, whch s usually mplemented va a lnea feedback shft egste FR In [], e output of e CUT s compessed usng anoe FR, whle n [] a nonlnea fnte state machne s used to obtan e ones countng n e output sequence When bo ccuts opeate n e test mode, a sequence of test pattens ae geneated appled to e CUT Fo e ones countng meodology, e ones count n e output sequence s ecoded en compaed w e pe-computed ones count of e fault-fee ccut usng e same nput pattens The dffeence,, s stoed Fo e sgnatue analyss technque, e fnal sgnatue s compaed to e pe-computed coect sgnatue If ey ae e same e test s sad to pass; oewse t fals Ths pocess of applyng pattens detemnng e value of, o pass/fal, s efeed to as a test sesson The pocess fo estmatng eo-ate conssts of cayng out test sessons, each usng a dffeent subset of test pattens fom e space of all test pattens In e followng, we wll concentate on e ones countng technque The analogous yet adcally dffeent analyss fo sgnatue analyss s gven n [] Cayng out ese test sessons esults n dffeences, denoted by,,, s It s assumed at all test pattens ae omly geneated at e total numbe of pattens, namely x, s a small facton of e nput patten space Thus ese numbes ae ndependent have e same dstbuton The sample mean sample vaance ae gven by e equatons Based on e sample mean vaance, e eo-ate estmato, ˆ, gven n [] s ˆ N N whee N n, n s e numbe of nput pns of e CUT Agan, N>> The selecton of depends on what objectve s to be satsfed In [], e auos analyzed e poblem of usng s ones countng technque to classfy chps accodng to e estmated eo-ate et epesent an eo-ate eshold value, usually specfed by an applcaton, ε a magn of eo allowed n e value of an estmated eo-ate, γ e confdence assocated w s estmaton If e estmated eo-ate, ˆ, of a CUT s smalle an, e CUT s accepted, oewse t s ejected The followng equements ae set fo e estmaton of eo-ate Fo a CUT w actual eo-ate, a If < ε, e pobablty of acceptng e CUT s geate an o equal to γ That s Pob ˆ < γ f < ε b If > ε, e pobablty of ejectng e CUT s geate an o equal to γ That s Pob ˆ > γ f > ε c If ε ε, e CUT may be accepted o ejected In [] e auos dscussed lowe bounds fo such at e above equements ae satsfed The analyss showed at e lowe bound fo e numbe of test pattens pe test sesson,, s, e lowe bound γ ε fo e numbe of test sessons,, s γ ε ε / ε Assume, ε γ977 The lowe bound of s about 97 e lowe bound of s about 7 Because e ones countng technque eques knowledge of e ones countng fo e fault-fee ccut n ode to compute, values must be stoed, whch esults n an exobtant amount of ovehead mployng e sgnatue analyss technque fo e same equements as above, only test pattens ae needed fo each of 58 test sessons These esults ae counte-ntutve fom e followng pont of vew In sgnatue analyss one only gets a bnay decson fom each test sesson, namely pass o fal Ignong e ssue of alasng, s decson s always coect, but e fal can be caused by,, up to eoneous esponses Hence a geat deal of nfomaton s lost pe falng test sesson On e oe h, fo ones countng, one detemnes an actual obseved ones countng, O, of,, *, Then e ecoded value s O O, at can take * on e values,,,,,,,,,, whee O s e ones countng fo e fault-fee ccut Hence ee appeas to be a much lage amount of nfomaton n an n pass/fal In fact, f, en fo sue e test has faled But f, e test may have passed o faled Thus alasng plays a bg ole n ones countng Howeve, t seems lke e value of gettng a non-bnay esult should moe an compensate fo e poblem of alasng In addton, f en no alasng s possble, so f e esponse s coect, f t s ncoect Hence, fo a lage enough value of, e eo-ate can be accuately estmated, us e lowe bound on s Thus t s useful to eexamne e analyss of eo-ate estmaton va ones countng To moe fully chaacteze e statstcal popetes of e estmato ˆ, ts mean vaance should be computed The mean of e estmato wll be close to e tue eo-ate when e estmato s based, equal to e tue eo-ate when e estmato s unbased The vaance of e estmato s not dectly calculated n [] Instead, t s mpled to be equal to e vaance of anoe om vaable namely e n [], whch s not always tue In s epot, we pesent an analyss of e ones countng technque at s dffeent fom e analyss povded n [] Ou analyss moe accuately descbes e chaactestcs of

4 UC TCHNICA RPORT CNG-8- ones countng technque fo e pupose of eo-ate estmaton We also compae s ones countng technque w e pevously publshed wok usng sgnatue analyss The compason shows at s ones countng technque s able to estmate eo-ate as effectvely as e sgnatue analyss technque, e hadwae oveheads fo bo technques ae compaable III TATITICA ANAI Consde a sngle output combnatonal ccut, C, a faulty veson of s ccut, C f In esponse to each nput patten, e output of C f can be classfed nto one of fou types, namely, /, /, / /, whee / means at e output of bo C C f ae ; / means at e output of bo C C f ae ; / means e at e output of C s C f s ; / means at e output of C s C f s If all possble N nput pattens ae appled to C f, a sequence of N outputs consstng of e fou types of outputs s geneated The eo-ate of C f s e ato of e numbe of outputs of type / plus type / to N ung a sngle test sesson, e obseved ones count s equal to e numbe of type / / outputs type x/ The esult s subtacted by e coect ones countng of C, whch s equal to e sum of type / / outputs of C f Ths dffeence, denoted by, epesents e dffeence between e ones count geneated by C f e ones count geneated by C It also equals e numbe of type / outputs mnus e numbe of type / outputs of C f A complete test conssts of test sessons Thus, ecoded ae numbes, each of whch s e value of fo a test sesson The eo-ate s estmated accodng to ese numbes Imagne at N possble outputs defne a collecton havng fou types of symbols In each test sesson, we choose outputs wout eplacement fom e collecton Fom e selected outputs, we do not ascetan e numbe of type / o / outputs, but e dffeence between e numbe of / / outputs Afte a test sesson s fnshed, we put ese outputs back nto e collecton test sessons ae caed out, esultng n numbes Fom ese numbes, we estmate e facton of outputs n e collecton at ae ee of type / o / et p be e facton of / outputs n e collecton, p e facton of / outputs n e collecton Thus, p p p, p ae all postve less an An oacle knows p p We wsh to estmate p p snce e estmated eo-ate equals e sum of e estmated values of p p Assume e outputs ae dawn one at a tme, we have a counte, ntalzed to If a type / output s dawn, e value of e counte s nceased by ; f a type / output s dawn, e value of e counte s deceased by ; f a type / o / output s dawn, e state of e counte s not changed The fnal state afte outputs ae chosen, e, a test sesson, s just e state of e counte, say Because outputs ae dawn omly, s a om vaable In e above pocess, t s mpled at we ae samplng wout eplacement Because N s assumed to be lage w espect to, whch s usually e case n pactce, e change n e facton of each type of output n e emanng collecton afte each of e outputs s selected s vey small can be gnoed o we wll teat s pocess as samplng w eplacement Thus, fo each dawng, e pobablty at e counte nceases by s p, e facton of / outputs; e pobablty at t deceases by s p, e facton of / output; e pobablty at t does not change s p p et X be a om vaable such at X w pobablty p, X w pobablty p, X w pobablty p p Thus X X X, whee X, X,, X ae dentcally ndependently dstbuted d om vaables w e same dstbuton as X Fom e pobablty densty functon PF of X, we see at xpectaton: { X p p aance: a{ X p p p p Thus e expectaton vaance of ae xpectaton: { { X X X { X p p aance: a { a{ X X X a{ X p p p p Fom e test pocess we obtan samples of e om vaable, namely,,, The sample mean e sample vaance ae defned by e equatons We ntend to estmate e two paametes, p p, n e dstbuton of A genec meod fo buldng an estmato s based on appoxmaton of e moments of e om vaable [] The fst ode moment of a om vaable s ts expectaton, whch s appoxmated by e sample mean The second ode moment s ts vaance, whch s appoxmated by e sample vaance Thus, we have { p p p a{ p p p p 5 Fom 5 we solve fo p p, obtan, hence p p Because e eo-ate s equal to e sum of p p, e estmated eo-ate, ˆ, s p

5 UC TCHNICA RPORT CNG-8-5 ˆ, 6 whch s also called e estmato of eo-ate ˆ s a functon of samples of o e estmato tself s a om vaable To evaluate e pefomance of an estmato, ts expectaton vaance need to be computed If e expectaton s equal to e tue value of e estmated quantty, e estmato s unbased; oewse t s based If e estmato s based, e dffeence between e expectaton of e estmato e tue value of e estmated quantty s of nteest malle dffeences mply bette estmatos The vaance of e estmato epesents how close e estmaton s to e expectaton of e estmato A lage vaance means e PF s somewhat flat e estmaton esult s lkely to be poo A small vaance mples at e PF s naow aound ts expectaton, e estmaton esult s lkely to be close to e expectaton of e estmato W e expectaton vaance of e estmato, we ae able to appoxmate e PF of e estmato If e type of dstbuton of e estmato s known, e PF of e estmato can be expessed explctly If e type of dstbuton s unknown, as t s fo e eo-ate estmato, e PF of e estmato s usually appoxmated by a nomal dstbuton, whch s a functon of e expectaton vaance of e estmato The pocedue fo devng e expectaton e vaance of e estmato ˆ s gven n Appendx A The expectaton of e estmato s { ˆ { p p [ p p p p ] 7 ts vaance s a { ˆ a{ / / 8 whee p p, p p, p p p 6 p p p The tue value of e eo-ate to be estmated s p p quaton 7 shows at e estmato s based Howeve, when s lage, e tems w can be gnoed e estmato becomes unbased ate we show at fo e poblems addessed n s epot, s lage, fo ese cases { ˆ p p We use e nomal dstbuton N ˆ { ˆ, a{ ˆ to appoxmate e dstbuton of e estmato Thus e PF of e estmato can be expessed as ˆ a{ ˆ P ˆ ˆ e 9 πa ˆ { We ae nteested n havng e estmated eo-ate be wn a cetan ange of accuacy, say [ ε, ε ], w confdence not less an γ, whee <ε << γ s between Thus, we eque ε ε ˆ a ˆ γ P ˆ ˆ ˆ ˆ d e d πa ˆ ε ε / a ε / a { ˆ { ˆ ε e π t / / a ˆ dt { Q ε whee e functon Q s defned as Q x see Appendx B Changng e fom of, we have Q γ / ε / a ˆ { x e t / / π dt a{ ˆ s a functon of pp, p p, Thus, ncludes sx quanttes, namely ε, γ,, p p,, whee ae unknown ε γ ae gven as pat of e test specfcatons To detemne values fo, whch ae test paametes fo cayng out eo-ate estmaton, some addtonal constants /o objectve functons ae needed Refeng back to ou I test poblem, some quanttes of nteests ae lsted next a nmze x, whch pmaly detemnes e total test tme; b nmze, whch pmaly detemnes e stoage cost fo e coect ones countng; c nmze c xc x, whch s e weghted cost fo bo test tme stoage cost, whee c c ae bo non-negatve cost coeffcent Refeng back to, s e tue eo-ate only known to an oacle, but we can guess a value fo efne ou guess once we have a value fo e estmato The quantty p p s also unknown, but agan we can attempt to appoxmate t Because e appoxmatons ae dffeent fo vaous stuatons, we wll deal w ese ssues n e next secton enttled case studes I CA TUI In [], e numbe of test pattens pe sesson,, has a lowe bound Howeve, ou analyss shows at any postve ntege s feasble fo Fst we consde two cases, namely / e, s vey lage Fom 8 we see at ae nvesely popotonal to each oe o esults n an uppe bound fo, / esults n a lowe bound of Then we consde e symmetc case of p p > At last we consde e geneal case In e followng except n ecton, we assume γ997 hence

6 UC TCHNICA RPORT CNG-8-6 Q γ / { ˆ ε / 9 Then educes to a Case : Fom 8, we have a{ ˆ 9 / 6 / / p p, we have p p p p p p w As p p Fom e defntons of,, we know at all of ese tems ae of e ode of o, 6 ae all of e ode of Keep tems w n e 9 denomnato gnoe tems w hghe odes of, we have a{ˆ can be ewtten as ε / 9 Replacng, w e functons of n leads to ε nce p p ae unknown, we cannot choose e value to be 9 6 / ε Howeve, we can choose to be e maxmum of 9 6 / ε Thus, 5 s satsfed t s guaantees at e estmated eo-ate s n e ange of [ ε, ε ] w confdence γ Fo /, whch s typcally e case, maxmzes 9 ε 6 o we choose to be 6 9 C ε, e, 6 Fo example, fo, ε5, we choose to be 65 When /<<, / maxmzes 9 ε 6 o can be chosen as 6 /, e, 95 ε be 565 Fo, ε5, we choose to Case : / Fo e case of /, not only do we assume at s vey lage, but also at >>, n whch case we gnoe all tems n 8 w an, obtan a{ˆ 7 Ignong e tem w eplacng w ts functon of, 7 educes to a { ˆ / Fom 7, we have 8 ε 8 mla to case, we can choose e value of to be e maxmum of 8 ε nce e maxmum value of 8 ε s 8 ε, we choose as C 8 ε 9 Fo example, fo ε5, we get C 7 us >> 7 Fom 9 we see at when s vey lage, e numbe of test sessons equed n eo-ate estmaton s ndependent of e eo-ate, only depends on e accuacy confdence of e estmaton Ths follows snce, f e eoate s extemely small, en e allowable eo n ou estmaton as specfed by ε allows ou test meodology to wok, fo lage values of eo-ate, ee s enough nfomaton gaeed by usng ese values of C to agan estmate e eo-ate If we choose 7 solve fo usng 9, we get Then e condton to appoxmate 8 w 7 s not satsfed In s case, we cannot use 9 to select Case Case epesent two exteme values of esultng n uppe lowe bounds fo Case : p p > When p p, e pobablty of obsevng a type / o / output ae e same Thus, e expected value of fo each sesson s zeo, e, e sample mean of,, s close to zeo In s case, t appeas at e ones countng test meodology becomes neffectve Howeve, when e estmato 6 educes to / Ths means at e eo-ate can be deved solely fom e vaance of W p p, we have Fom 8, we have a { ˆ

7 UC TCHNICA RPORT CNG-8-7 Assume s lage When compaed w /, / can be gnoed When compaed w /, we can gnoe /, /, / / Thus, educes to { ˆ a / Fom, we have / ε /, e, 9 9 C ε Any postve ntege s legtmate fo When, 9 / ε When s vey lage, 8/ ε These esults fo ae consstent w ose deved fo Cases Fg shows e elatonshp between fo dffeent eo-ates ε 5 ognumbe of Test essons Fom top to bottom, e cuves coespond to e cases of, 5,, Numbe of Test ectos pe esson Fg vs based on q fo dffeent eo-ates ε 5 The ponts maked by an X coespond to e selecton of at mnmze x Consde e case of a sngle stuck at fault n a sngleoutput XOR ccut at causes half of all outputs to be wong Among ose eoneous outputs, half ae of type / e oe half of type / o p p / / Ths povdes an example of Case If e output lne s stuck-at, en agan / but now all eos ae of type / To see how e vaance s nstumental n detemnng e estmated value of eo-ate, agan consde a ccut such at a omly selected half of all possble nput pattens map nto, e oe half map nto Now consde a faulty veson of s ccut, whee agan a omly selected half of all possble nput pattens map nto, e oe half map nto o e tue eo-ate of e faulty ccut s / Now, fo bo e good faulty ccuts, e aveage value of e ones countng s / Thus, e expected value of e dffeence of ones countng between good ccut faulty ccut s zeo In addton, fo e faulty ccut, e pobablty of obsevng a /, /, / o / type esponse s /, e, p p / Assume Fo each sesson, e possble values of ae,,,,,,, If fo a sesson, en all fou outputs ae / type Thus, e pobablty of s / /56 If fo a sesson, en ee outputs ae / type one output s ee / o / type Thus, e pobablty of s x/ x/ / mlaly, we can compute e pobablty of beng,,,,, o As a esult, e pobabltes of beng,,,,,,, ae /56, /, 7/6, 7/, 5/8, 7/, 7/6, / /56, espectvely W e dstbuton functon of, we have e expectaton of to be zeo e vaance to be Fom e expectaton vaance of, we know at e sample mean of,, s about zeo e sample vaance of,, s about Fom e estmato / /, we obtan e estmated eo-ate to be /, whch matches e tue eoate Geneal Case Fo e geneal case, we make no assumptons of e values of p, p γ Howeve, we assume s lage To make ou dscusson clea, we show a copy of 8 below a{ ˆ a{ / / Because s lage, e tem a / can be gnoed when compaed to a / Fo e same eason, e tems /, /, a, a / a / can be gnoed when compaed to / Thus, e vaance of e estmato becomes a{ ˆ w e functons of Replacng,, we have a { ˆ 6 Then, becomes hence 6 ε / Q γ /

8 UC TCHNICA RPORT CNG-8-8 Q γ / 6 5 ε Wout knowng e value of, we choose to be e maxmum achevable value of ght h sde of 5 It can be shown see Appendx C at when s less an /5, whch s geneally e case of nteest, esults n 5 beng maxmal Thus we choose as Q γ / 6 ε Fo γ 997, 6 educes to Ths s expected because Case, fom whch s deved, s a specal case of Case Fo, 6 educes to 6 Fo /, 6 educes to 9 These esults mply at e analyss of e geneal case s consstent w e analyss of ts specal cases ettng allows us to fnd a lowe bound on Fo γ997, s lowe bound s gven by 9 Fo ε, e lowe bound s 8 Usually ε s much smalle an, as ε deceases e lowe bound on nceases Thus n 8, whee an tem exst, t s appopate to gnoe tems contanng,,,, Ths justfes e appoxmatons used to obtan fom 8 Now consde mnmzng test tme whch s popotonal to x Assume γ997 Fom 6, we have 9 When /, en ε 9 6 / ε x has a mnmal value of 9/ / ε When s small, 6/ ε x mn 8 / ε In Fg, e ponts maked by X coespond to e selectons of at mnmze x It can be seen at fo dffeent values of eo-ate, e values of ae almost e same, namely 6/ ε Fo eo-ate estmaton usng sgnatue analyss [], t s ecommended to set /, whch leads to 5/ ε 5 / ε Thus we see at e ones countng technque fo eo-ate estmaton s compaable to sgnatue analyss n tems of total test tme, a lttle hghe n tems of ovehead cost IUATION In e above analyss, we appoxmated e pobablty densty functon of e estmato w e nomal dstbuton N ˆ { ˆ, a{ ˆ Then we developed a way to select to satsfy e accuacy confdence equement of eoate estmaton based on s appoxmaton In s secton, we descbe ou esults of estmatng e eo-ate va smulaton By epeatng e smulaton pocess many tmes, we can collect a lage numbe of estmated eo-ates compae e dstbuton w N ˆ { ˆ, a{ ˆ The smulaton s mplemented as follows A om numbe geneato geneates ee numbes, namely a w pobablty p, w pobablty p w pobablty p p The numbe of s e numbe of s n a sequence of geneated numbes ae counted sepaately The dffeence s ecoded Ths epesents a test sesson test sessons ae caed out, numbes ae ecoded W e estmato pesented n ecton, e eo-ate p p s estmated The above pocedue s epeated tmes esults n eo-ate estmatons The dstbuton of e estmated eo-ates s geneated We use e ATAB tool nomplot to detemne f s data s consstent w a nomal dstbuton Nomplot dsplays e cumulatve dstbuton of e data In e plot, a supemposed lne s dawn to ft e sample data If e data ae nomally dstbuted, e plot appeas lnea Fst, we set p 6, p, 5 Fg a shows e dstbuton of e data t appeas to be stbuton Pobablty ata fom mulaton a Nomal Pobablty Plot 6 8 ata b Fg a stbuton of estmated eo-ate data fom smulaton b The output fom ATAB tool nomplot to test whee e data ae nomally dstbuted Fo s fgue, p 6, p, 5

9 UC TCHNICA RPORT CNG-8-9 nomal The output of nomplot s shown n Fg b, s faly lnea, confmng at e data has a nomal dstbuton Now, a nomal dstbuton can be defned by ts mean stad devaton Fom e data, we estmate e mean to be e vaance to be -6 o e set of eo-ate data has nomal dstbuton N, - In ecton, we used a nomal dstbuton to appoxmate e dstbuton of e estmato The mean of e estmato s gven by 7, esults n a value of The vaance of e estmato s gven by 8, esults n a value of -6 o n ou analyss, we would use N, -6 to be e dstbuton of e estmato Thus ou analytcal esults closely match e smulaton esults Next, we choose dffeent values of whle keepng p 6 p Fg shows e dstbuton of e estmated eo-ate data fom smulaton e nomal dstbuton test fom nomplot fo e case whee s about tmes / Fo ε, e lowe bound on s 8 s close to e lowe bound on, us e case outlned n ecton s satsfed The 5 fgue shows at e smulaton data ae nomally dstbuted Fom e data, we estmate ts nomal dstbuton to be N, 5-7 Fom e analyss n ecton, we obtan e dstbuton of e estmato to be N, 8-7, whch agan s an excellent match Fnally, we consde e case whee p p 5, 5 nce s small, one would expect at fo most test sessons few f any eos would occu In s case, e estmated eo-ate s manly deved on e vaance of e sample data as mentoned n ecton The dstbuton of eo-ate data fom smulaton e output of nomplot ae dsplayed n Fg 5, whch shows e smulaton data has a nomal dstbuton The dstbuton functon of e data s estmated to be N, 985-8, whch matches e dstbuton of e estmato fom analyss, namely N, I CAIFING CHIP IA THIR RROR-RAT One applcaton fo eo-ate estmaton s to assgn chps stbuton 5 stbuton 8 6 Pobablty 6 8 ata fom smulaton a Nomal Pobablty Plot ata b Fg a stbuton of estmated eo-ate data fom smulaton b Output fom ATAB tool nomplot to test whee data ae nomally dstbuted Fo s fgue, p 6, p, Pobablty 6 8 ata fom smulaton a Nomal Pobablty Plot ata b Fg 5 a stbuton of estmated eo-ate data fom smulaton b Output fom ATAB tool nomplot to test whee data ae nomally dstbuted Fo s fgue, p 5, p 5, 5

10 UC TCHNICA RPORT CNG-8- to bns at coespond to eo-ate anges at ae defned by eshold eo-ate values Recall at a eshold sepaates a ange nto two adjacent sub-anges Theshold dvdes e doman of eo-ates nto two ange,,, consequently falng chps ae pattoned nto two types, A B The eo-ate of a type A chp s equal to o less an, whle at of a type B s geate an Testng classfes a chp nto type A o type B accodng to e estmated eoate ˆ Namely, f ˆ <, e chp s classfed as type A; oewse, t s classfed as type B Unfotunately, e om vaable ˆ can be geate an even ough e tue eoate,, s smalle an vce vesa The chance fo s to occu nceases apdly as ˆ appoaches e value o e test can classfy chps ncoectly In statstcs such a test s called a hypoess test In ou case, e two hypoeses ae: H: The chp s type B, e, > ; H: The chp s type A, e, Ths test geneates fou possble outcomes The chp s type B, classfed as type B; e chp s type B, classfed as type A; e chp s type A, classfed as type A; fnally e chp s type A classfed as type B Outcome esults n a lowe pce chp sold eoneously at a hghe pce, whle outcome esults n a hghe pce chp sold at a lowe pce Outcome s lkely acceptable to customes outcome s not o e test should lmt e pobablty of e occuence of outcome Assume t s equed at e pobablty of any type B chp beng classfed as type A be smalle an β, whee β<< Accodng to e analyss of eo-ate estmaton, e estmated eo-ate has nomal dstbuton w ts expectaton beng e tue eo-ate Fg 6 shows e dstbuton of estmated eo-ate of a chp whose tue eoate s a chp whose tue eo-ate s geate an It can be seen at fo a chp w tue eo-ate geate an, e fue e tue eo-ate s fom, e lowe at pobablty of outcome occung Ths pobablty s epesented by e dash aea unde e cuve Howeve, when e tue eo-ate s equal to, e pobablty of outcome s always 5% Thus, e equement s neve satsfed To solve s ssue, we defne anoe eshold, namely n, at s smalle an, postulate e followng classfcaton cteon: a b Fg 6 a PF of estmated eo-ate of a chp whose tue eo-ate s b PF of estmated eo-ate of a chp whose tue eo-ate s geate an If e estmated eo-ate of a chp s smalle an n, t s classfed as type A If e estmated eo-ate of a chp s equal to o geate an n, t s classfed as type B Assume at e pobablty of outcome occung s stll lmted to β Thus s constant also holds when, hence e pobablty of outcome s smalle an β when > o we eque β Pob ˆ < n When, e estmated eo-ate has nomal dstbuton N, a{ˆ Thus, Pob ˆ < n Q n a{ ˆ n Q β β, whch s equvalent to a { ˆ Fom, we aleady have { 6 a ˆ Thus n Q β n 6 7 o equvalently, Q β 6 8 n 8 descbes e equement fo such at e pobablty of outcome occung s lmted to β Wout knowng e value of, we choose to be geate an e maxmum value of e ght h sde of 8 W beng smalle an, e ght h sde of 8 s maxmal at e value Q β when o we n choose accodng to e expesson Q β 9 In 9, β ae gven n e test specfcaton We need to detemne e value of, n Fom ou pevous analyss, e selecton of depends on what cost functon s mnmzed Assume has been detemned Then s detemned by n As n deceases, deceases Because smalle values of esult n less stoage cost, t s mpotant to keep small Howeve, smalle n causes moe type A chps to be classfed as type B, meanng a loss n poft o ee s a tadeoff between stoage cost poft Fo a type A chp, ts tue eo-ate,, may be smalle o geate an n Gven, e pobablty of outcome can be computed Consde e case of < n The estmated eo-ate has nomal dstbuton 6 N The pobablty of,

11 UC TCHNICA RPORT CNG-8- outcome, e e estmated eo-ate beng geate an n s n Q / 6 When <, takes on a maxmal value of n Q, whch s e uppe bound of / e pobablty of outcome fo e case of < n When n < <, e pobablty of outcome s n Q / 6 nce p p,, t can be shown at f < / en s maxmal when has e value n Q / 5 6 mlaly, e pobablty of outcome fo e case of > can be computed Table summazes e uppe bounds on e pobablty of eoneous classfcaton fo ese ee cases mla to e eo-ate estmaton technque based on sgnatue analyss [], t s not necessay to apply all test sessons befoe makng a decson because e fae e tue eo-ate away fom e eshold, e less pobablty of makng wong classfcaton The numbe of test sessons s based on e assumpton of e wost case, e, when Howeve, a defectve chp usually does not epesent e wost case, sometmes neve epesents such a case The classfcaton equement s at e pobablty of an estmated eo-ate less an n be smalle an β f e tue eo-ate s geate an o t s possble to make a decson wout executng all test sessons as long as e pobablty of makng a wong decson s smalle an β et ms be e mnmal numbe of test sessons equed fo a chp w eo-ate at satsfes e constant mposed by β Fo >, e uppe bound on e pobablty at e estmated eo-ate s smalle an n s n Q Fo < n, e uppe bound of / e pobablty at estmated eo-ate s geate an n s n Q Bo fomulas ae lsted n / Table Then ms should satsfy n Q β f >, / ms ms n Q β f < n / ms ms TAB I TH AXIA IKIHOO OF AKING AN RRONOU CAIFICATION Tue eo-ate < n, estmated eo-ate ˆ > n Type A classfed as type B n << ˆ > n Type A classfed as type B ˆ < n > Type B classfed as type A Uppe Bound of e Pobablty of akng oneous Classfcaton Q Q n / When n < <, ms s equal to as specfed by 9 As an example, let n 9, β 5 s chosen to be / Fg 7 shows e mnmal numbe of test sessons ms fo dffeent tue eo-ates As moves away fom e ange of [ n, ], ms quckly deceases o fo eo-ates fa fom [ n, ], only a small numbe of test sessons ae needed To be able to make ealy decson befoe applyng all test sessons, e test pocedue must be modfed An ognal test s dvded nto multple phases, each of whch conssts of a dsjont subset of e test sessons Afte each test phase s completed, e eo-ate s estmated based on e esults fom e test sessons appled so fa, assumng e estmated eo-ate s e tue eo-ate If e estmated eo-ate s smalle an n, e pobablty of makng an eoneous classfcaton s calculated usng If e estmated eo-ate s geate an, e pobablty of makng an eoneous classfcaton s calculated usng If e computed pobablty s smalle an β, e test stops, oewse t contnues If e estmated eo-ate s n e ange [ n, ], e test also contnues unless all test sessons nmal numbe of test sessons, ms n 5 / n Q 6 5 Tue eo-ate, Fg7 The mnmal numbe of test sessons fo dffeent tue eo-ates

12 UC TCHNICA RPORT CNG-8- have been un, n whch case testng s fnshed II XPRINTA RUT To valdate s chp classfcaton technque, we appled t to e ICA 85 benchmak ccut C at has 7 pmay outputs Because ou technque s cuently only applcable to sngle output ccuts, we ceated seven sngle output ccuts fom C, labeled as C_, C_, C_, C_, C_5, C_6 C_7 The seven ccuts have e same netlst as C Fo each of ese netlsts, only one pmay output of C s teated as e output of e new ccut oe outputs ae teated as ntenal wes Fo example, e output pn of C s e output of C_, output pn,,, 5, 6 7 of C ae teated as ntenal wes n C_ To model a defect, we used e sngle stuck-at fault model ach of ese ccuts has 86 sngle stuck-at faults Thus, coespondng to each fault-fee ccut, ee ae 86 faulty ccuts Consde e 86 faulty ccuts of C_7 nce we know e actual faults n e ccut, we can obtan e actual eo-ates see Fg 8 To classfy ese ccuts, we set, n 9, β5 5 The eo-ate of a type A ccut s n e ange,, fo a type B ccut [, Fom 9, e maxmal numbe of test sessons,, s 8 We use e 5 mult-phase test scheme descbed n e pevous secton fo classfcaton In e fst phase, * test sessons ae executed The collected data s not statstcally meanngful f * s too small In e followng phases, only one test sesson s appled ven ough moe test sessons can be appled, we choose one because we want to fnd e exact numbe of e test sessons needed Afte each phase, e estmated eo-ate s computed e condton fo stoppng s checked The total numbe of test sessons fo each ccut s ecoded Fg 9 shows e hstogam of test sessons fo all e ccuts Fom Fg 9, t s seen at fo many ccuts, only a small numbe of test sessons ae needed Ths means at e eoate fo each of ese ccuts s fa away fom e ange 9, Ths esult s consstent w Fg 8, at shows at only a small facton of e ccuts have an eo-ate among nea e ange 9, To fue demonstate e coelaton between e numbe of test sessons eo-ate, n Table we lst e aveage numbe of test sessons fo dffeent sub-anges of eo-ate Fo eo-ate fa fom 9~, e aveage numbe of test sessons s small As e eo-ate gets close to 9~, e aveage numbe of test sessons nceases Ths s consstent w e analyss n ecton 6 Fo eo-ates n e ange 9~, e aveage numbe of test sessons s 995, whch s not equal to e maxmal numbe of test 8 stbuton 5 stbuton 6 stbuton o-ate a o-ate 5 b Fg8 Nomalzed hstogam of eo-ates of 86 faulty ccuts assocated w C_7 a The eo-ate ange s fom to b The eo-ate ange s fom to 55 Note: e scales of a b ae dffeent stbuton Test essons a 6 8 Test essons b Fg9 Nomalzed hstogam of e numbe of test sessons appled when classfyng e 86 faulty ccuts of C_7 a The numbe of test sessons s n e ange of to b The numbe of test sessons s n e ange of to Note: e scales of a b ae dffeent

13 UC TCHNICA RPORT CNG-8- sessons 8 Ths s because fo some ccuts, e estmated eo-ate s ee below 9 o above, stoppng condton s satsfed befoe unnng 8 test sessons TAB II ARAG NUBR OF TT ION OF FAUT C_7 CIRCUIT FOR IFFRNT RROR-RAT RANG o-ate Range The pecentage of ccuts msclassfed as type A s called test escape, ose msclassfcaton as type B s called yeld loss Intutvely, f e eo-ate of a ccut s fa fom e eshold, t s less lkely to be msclassfed Because most of e 86 ccuts n ou expement have an eo-ate fa fom e eshold, e test escape yeld loss should be low Table lsts e numbe of msclassfcaton fo ee dffeent eo-ate anges Thus fo s expement e test escape s /7% e yeld loss s /9% The same classfcaton expement was appled to benchmak ccut C88 A sngle output ccut, namely C88_6, was deved fom C88 such at one pmay output of C88 s e only output of C88_6, e oe pmay outputs of C88 ae teated as ntenal wes of C88_6 Because C88_6 has 96 sngle stuck-at faults, we obtaned 96 faulty copes of C88_6 ach faulty copy coesponds to a sngle stuck-at fault In e expement,, n 9, β5 5 Table lsts e aveage numbe of test sessons fo dffeent sub-anges of eo-ate e numbe of faulty ccuts n each ange mla to e esults fo C_7, e numbe of test sessons nceases when e eo-ate ange s close to 9~ Table also quantfes e numbe of msclassfcatons made Of e 55 type A ccuts, 7 ae msclassfed Of e 6 type B ccuts, 6 ae msclassfed The esultng yeld loss s 5% e test escape s 9% The esults of ese expements on C_7 C88_6 ae consstent w ou analytcal esults When e eo-ate of a ccut s fa fom e eshold, coect classfcaton o-ate Range Aveage Numbe of Test essons TAB III ICAIFICATION IN TH XPRINT FOR C_7 Actual Type Actual Numbe of Faulty Ccuts Numbe of Ccuts n e Range ~ 77 ~8 98 8~ 8 7 ~ ~7 9 7~ ~ ~ 5 9 ~ 59 8 ~ ~ 56 ~5 99 5~ 5 7 Numbe of sclassfed Ccuts ~9 Type A 55 9~ Type A 6 9 ~ Type B 9 decson can be made afte a small numbe of test sessons As e eo-ate of a ccut appoaches e eshold, moe test sessons ae needed In ou expements, we choose β to be 5, whch lmts e pobablty of msclassfcaton, use β fo e wose stuaton to detemne oe test paametes When e eo-ate of a ccut s fa fom e eshold, e pobablty of makng a wong decson s actually smalle an β Fom table t s seen at e pobablty of makng a wong decson nceases as e actual eo-ate appoaches e eshold o-ate Range TAB I XPRINTA RUT FOR FAUT C88_6 CIRCUIT Actual Type Aveage Numbe of Test essons Numbe of Actual Ccuts n e Range Numbe of sclassfed Ccuts ~ A 97 ~8 A ~ A 6 7 ~5 A ~7 A 7 7 7~9 A ~ A ~ B ~ B 7 ~7 B ~ B 68 7 ~5 B ~ B 9 9 In e expement fo C88_6, 57 type A ccuts ae coectly classfed as type A, 6 type B ccuts ae msclassfed as type A Thus, 55 ccuts ae classfed as type A w 6 of em actually beng type B If we sell all 55 ccuts, n pncple 6 mght justfable be etuned The facton of chps at ae beyond-toleance FBT, namely 6/55, should not be confused w e classcal noton of defects pe mllon P used as a measue fo poduct qualty We eque at e maxmal pobablty of msclassfcaton fo a ccut be β Theoetcally, fo e wost scenao n classfcaton, e numbe of actual type B ccuts at ae msclassfed as type A s β e numbe of actual type B ccuts; e numbe of actually type A ccuts at ae msclassfed as type B s β e numbe of actual type A ccuts Thus e numbe of ccuts at ae classfed as type A s e numbe of actual type A ccuts β e numbe of actual type A ccuts β e numbe of actual type B ccuts Then, FBT s equal to β e numbe of actual type B ccuts e numbe of ccuts at ae classfed as type A In pactce, e pobablty of msclassfcaton s lowe an β when e eo-ate of a ccut s fa fom e eshold Thus, e eal value of FBT s usually lowe en ts eoetcal value In ou expement on e 96 faulty ccuts of C88_6, e expemental value of FBT s 6/558%, whle e eoetcal value of FBT s 5x6/55-5x555x6 o 696%

14 UC TCHNICA RPORT CNG-8- III CONCUION o-toleance mates new domans fo I tests, such as eo-ate Hee, e objectve s to classfy chps accodng to e eo-ate ae an pass/fal, o X Hz vs Hz The fundamental task assocated w s classfcaton s e estmaton of eo-ate An eo-ate estmaton technque based on sgnatue analyss was pesented n [] [], whee t was shown at s estmaton chp classfcaton can be effectvely mplemented w easonable BIT test esouce Anoe eo-ate estmaton technque based on ones countng also esulted n an estmato classfcaton technque [] The statstcal popetes of s estmato wee not fully studed, some esults wee non-ntutve, e equed esouces needed to cay out e test pocess appeaed to be lage an necessay In s epot, we agan consde e poblem addessed n [], develop a maematcally smple estmato In addton, we analyze e statstcal chaactestcs of e estmato, such as ts expectaton, vaance pobablty densty functon We detemne e condtons when s estmato s based vs unbased Based on e statstcal analyss, we descbe a meod fo selectng e values fo two key test paametes, namely e numbe of test pattens pe sesson,, e numbe of test sessons, These paametes mpact e test esouces used fo eo-ate estmaton We show how ese paametes ae a functon of bo test tme stoage equements In addton, we povde useful pactcal bounds fo ese two paametes The esults of ou analyss show at e test esouces equed fo eo-ate estmaton based on ones countng ae qute compaable to ose needed when sgnatue analyss s employed Thus, e majo dffeence n ese two appoaches s at sgnatue based eo-ate analyss can opeate on a mult-output ccut, whle ones countng pocesses one output at a tme unless e test hadwae s eplcated fo each output lne We also addess e poblem of classfyng chps based on e eo-ate The poposed classfcaton pocedue s pattoned nto multple phases Ths pocess can sgnfcantly educe classfcaton tme wout any loss n e qualty of classfcaton Thoughout s epot seveal assumptons have been made, such as assumng a nomal dstbuton fo some om vaables, doppng tems at lead to second o d ode effects We have addessed ese ssues n ou expements, valdated at ese assumptons ae appopate We have also consdeed vaous bounday o exteme condtons, such as whee e test leng of a test sesson s o vey lage In ou analyss, we assume all possble nput pattens ae equally lkely to appea have unfom dstbuton Thus e test patten geneato n BIT s mplemented w FR geneates pseudo om test pattens Fo some ccuts, e functonal nput may have dffeent statstcs an unfom dstbuton To deal w s stuaton, test pattens can be obtaned fom functonal test ccut nstead of FR, e same pocedue s appled Thus, e estmated eoate s a condtonal pobablty of outputtng an eo gven e eal statstcs of nput pattens Thus, f e dstbuton of functonal nput pattens s known, t may be feasble to eplace e PRPG w one at poduces pattens whose dstbuton moe closely mmcs e eal wold Fo multple output ccuts, e output lnes may not be of e same mpotance Thus, when ese output lnes ae consdeed ndependently, each of ese lnes wll be assocated w a dffeent eo-ate eshold Fo example, e most sgnfcance bt lne mght have a much smalle eoate eshold an e least sgnfcance bt lne Unde s stuaton, e technque pesented n s epot can be used to estmate e eo-ate of each output lne justfy whee e eo-ate s smalle an a eshold When e output lnes of a multple output ccut ae consdeed as a whole fo eo-ate, ts eo-ate cannot be smply deved fom e eo-ate of each output lne because of e coelaton of eos among dffeent output lnes Ths poblem has been solved by usng sgnatue analyss [], whle usng ones countng to solve s poblem s e bass of a focomng pape APPNIX A xpectaton aance of e stmato Fo eo-ate estmaton based on ones countng compesson, we popose usng e estmato ˆ, whee ae, espectvely, e sample mean sample vaance of e sampled om vaable In s appendx, we deve expessons fo e expectaton vaance of e estmato Pelmnaes In ecton, we defned X, whee X, X,, X ae d have e same dstbuton as e om vaable X The PF of X s PobXp, PobX p, PobXp p, whee p, p p p To smplfy futue maematcal expessons, we defne,, as follows: X p, { X X p { p p p p p { { X X p, X p p p p p p p, p

15 UC TCHNICA RPORT CNG-8-5 { X { X X X X X p p 6 p p s e expectaton of X,, ae e nd, d cental moments of X Addtonal symbols ae defned below {,,,,, { {, { {, { {, { { denotes e expectaton of,,, ae d have e same dstbuton as,, ae e st, nd, d ode moments of nce X n tems of,, { { X { a, we ae able to expess,, { { p p { X X X { { X { X { X X X { X { { X X { evaton of e xpectaton aance of e stmato The expectaton of e sample vaance of a om vaable s equal to e expectaton of e om vaable [] o we have { a{ A The expectaton of s computed as follows { j j, j { { { { { a{ { A Thus, e expectaton of e estmato s { { { a{ a{ { / / / p p [ p p p p ] A Next we compute e vaance of e estmato a a a Cov, a{ {, a Cov A We next compute each component n A Now { { { a A5 whee { a{ { { A6 Thus, { 6 a A7 a et We next consde e tem { / Then

16 UC TCHNICA RPORT CNG-8-6 Recall at { { { a { { { a A8 Hence, { { { A9 In A9, { A { { A { { A Theefoe, { A o fnally fo A8 we have { { a A We now consde e tem { { {, Cov n A, whee { A5 { { { { a A6 Theefoe,, Cov A7 ubsttutng { a, { a, Cov nto A usng A7, A A7, e vaance of e estmato s obtaned as a A8 Because s lage compaed to, we assume at Now we have a A9 B Q Functon: Qx Q functon Qx computes e ght tal pobablty of nomal dstbuton N, o Q functon s defned as x t dt e x Q / π Usng Q functon, many pobablty elated to a om vaable havng nomal dstbuton N, σ can be smply expessed smply Hee ae some examples et T denote such a om vaable Pobablty of T beng n x, s x t dt e x Q / / σ πσ σ Pobablty of T beng n -, x s - Qx-/σ Pobablty of T beng n -,, whee >, s - Q/σ C Justfcaton of 5 w </5 beng axmzed when n ecton quaton 5 n ecton states at Q 6 / ε γ

17 UC TCHNICA RPORT CNG-8-7 Because p p p p, we have < < Thus s maxmzed when o s The tem 6 T can be ewtten as a 5 / 6 7 / 6 T 6 C When > 5 / 6, T deceases as nceases Fo </5, 5 / 6< On e oe h, s always nonnegatve Fo </5, T s maxmzed when Thus, fo </5, bo 5 / 6 7 / 6 6 a ae maxmzed when Hence, fo </5, Q γ / 6 ε s maxmzed when, ts maxmum s Q γ / ε C [] Z Pan A Beue, stmatng o Rate n efectve ogc Usng gnatue Analyss, I Tans on Computes, ol 56, No 5, pp 65-66, ay 7 [] hahd K Gupta, stmatng o Rate ung elf-test va One s Countng, Poc Int l Test Conf, pape 5, 6 [] G Casella R Bege, tatstcal Infeence, uxbuy Pess, nd ed, [5] A Beue H Zhu, o-toleance ult-eda, Poc Int l Conf on Intellgent Info Hdng ultmeda, pp 5-5, 6 [6] Abamovc, A Beue A Fedman, gtal ystems Testng Testable esgn, Wley-I Pess, 99 [7] N K Jha K Gupta, Testng of gtal ystems, Cambdge Unv Pess, st ed, ACKNOWGNT The auos ank hdeh hahd Pofessos eep Gupta, Ke Chugg Antono Otega fo e valuable suggestons on s wok ome deas mpovements n s epot ae attbutable to em RFRNC [] Intenatonal Technology Roadmap fo emconductos: eld nhancement, 6 Update, [] Butts, A ehon C Goldsten, olecula lectoncs: evces, ystems Tools fo Ggagate, Ggabt Chps, Poc Int l Conf on Compute Aded esgn, pp -, Nov -, [] R I Baha, Tends Futue ectons n Nano tuctue Based Computng Fabcaton, Poc Int l Conf on Compute esgn, Oct -, 6 [] I Koen Z Koen, efect Toleance n I Ccuts: Technques eld Analyss, Poc I, ol 86, No 9, pp 89-88, ept 998 [5] A Beue, K Gupta T ak, efect o Toleance n e Pesence of assve Numbes of efects, I esgn Test of Computes, pp 6-7, ay-june [6] A Beue, Intellgble Test Technques to uppot o-toleance, Poc Asan Test ymp, pp 86-9, [7] H H Kuok, Audo Recodng Appaatus Usng an Impefect emoy Ccut, U Patent 5758, Patent Tademak Offce, 995; [8] P Agade, gtal ecetay, U patent 56555, Patent Tademak Offce, 997; [9] H Chung A Otega, Analyss Testng fo o Toleant oton stmaton, Poc I Int l ymp on efect Fault Toleance n I ystems, pp 5-5, Nov 5 [] H Chung A Otega, ystem evel Fault Toleance fo oton stmaton, Techncal Repot UC-IPI5, gnal Image Pocessng Insttute, Unv of ouen Calfona, [] A Beue, stmatng o Rate n o Toleant I Chps, I Int l Wokshop on lectonc esgn, Test Applcatons, pp - 6, Januay

Multistage Median Ranked Set Sampling for Estimating the Population Median

Multistage Median Ranked Set Sampling for Estimating the Population Median Jounal of Mathematcs and Statstcs 3 (: 58-64 007 ISSN 549-3644 007 Scence Publcatons Multstage Medan Ranked Set Samplng fo Estmatng the Populaton Medan Abdul Azz Jeman Ame Al-Oma and Kamaulzaman Ibahm

More information

A Brief Guide to Recognizing and Coping With Failures of the Classical Regression Assumptions

A Brief Guide to Recognizing and Coping With Failures of the Classical Regression Assumptions A Bef Gude to Recognzng and Copng Wth Falues of the Classcal Regesson Assumptons Model: Y 1 k X 1 X fxed n epeated samples IID 0, I. Specfcaton Poblems A. Unnecessay explanatoy vaables 1. OLS s no longe

More information

P 365. r r r )...(1 365

P 365. r r r )...(1 365 SCIENCE WORLD JOURNAL VOL (NO4) 008 www.scecncewoldounal.og ISSN 597-64 SHORT COMMUNICATION ANALYSING THE APPROXIMATION MODEL TO BIRTHDAY PROBLEM *CHOJI, D.N. & DEME, A.C. Depatment of Mathematcs Unvesty

More information

The Greatest Deviation Correlation Coefficient and its Geometrical Interpretation

The Greatest Deviation Correlation Coefficient and its Geometrical Interpretation By Rudy A. Gdeon The Unvesty of Montana The Geatest Devaton Coelaton Coeffcent and ts Geometcal Intepetaton The Geatest Devaton Coelaton Coeffcent (GDCC) was ntoduced by Gdeon and Hollste (987). The GDCC

More information

Optimization Methods: Linear Programming- Revised Simplex Method. Module 3 Lecture Notes 5. Revised Simplex Method, Duality and Sensitivity analysis

Optimization Methods: Linear Programming- Revised Simplex Method. Module 3 Lecture Notes 5. Revised Simplex Method, Duality and Sensitivity analysis Optmzaton Meods: Lnea Pogammng- Revsed Smple Meod Module Lectue Notes Revsed Smple Meod, Dualty and Senstvty analyss Intoducton In e pevous class, e smple meod was dscussed whee e smple tableau at each

More information

3. A Review of Some Existing AW (BT, CT) Algorithms

3. A Review of Some Existing AW (BT, CT) Algorithms 3. A Revew of Some Exstng AW (BT, CT) Algothms In ths secton, some typcal ant-wndp algothms wll be descbed. As the soltons fo bmpless and condtoned tansfe ae smla to those fo ant-wndp, the pesented algothms

More information

A. Thicknesses and Densities

A. Thicknesses and Densities 10 Lab0 The Eath s Shells A. Thcknesses and Denstes Any theoy of the nteo of the Eath must be consstent wth the fact that ts aggegate densty s 5.5 g/cm (ecall we calculated ths densty last tme). In othe

More information

An Approach to Inverse Fuzzy Arithmetic

An Approach to Inverse Fuzzy Arithmetic An Appoach to Invese Fuzzy Athmetc Mchael Hanss Insttute A of Mechancs, Unvesty of Stuttgat Stuttgat, Gemany mhanss@mechaun-stuttgatde Abstact A novel appoach of nvese fuzzy athmetc s ntoduced to successfully

More information

Tian Zheng Department of Statistics Columbia University

Tian Zheng Department of Statistics Columbia University Haplotype Tansmsson Assocaton (HTA) An "Impotance" Measue fo Selectng Genetc Makes Tan Zheng Depatment of Statstcs Columba Unvesty Ths s a jont wok wth Pofesso Shaw-Hwa Lo n the Depatment of Statstcs at

More information

Scalars and Vectors Scalar

Scalars and Vectors Scalar Scalas and ectos Scala A phscal quantt that s completel chaacteed b a eal numbe (o b ts numecal value) s called a scala. In othe wods a scala possesses onl a magntude. Mass denst volume tempeatue tme eneg

More information

Chapter Fifiteen. Surfaces Revisited

Chapter Fifiteen. Surfaces Revisited Chapte Ffteen ufaces Revsted 15.1 Vecto Descpton of ufaces We look now at the vey specal case of functons : D R 3, whee D R s a nce subset of the plane. We suppose s a nce functon. As the pont ( s, t)

More information

If there are k binding constraints at x then re-label these constraints so that they are the first k constraints.

If there are k binding constraints at x then re-label these constraints so that they are the first k constraints. Mathematcal Foundatons -1- Constaned Optmzaton Constaned Optmzaton Ma{ f ( ) X} whee X {, h ( ), 1,, m} Necessay condtons fo to be a soluton to ths mamzaton poblem Mathematcally, f ag Ma{ f ( ) X}, then

More information

8 Baire Category Theorem and Uniform Boundedness

8 Baire Category Theorem and Uniform Boundedness 8 Bae Categoy Theoem and Unfom Boundedness Pncple 8.1 Bae s Categoy Theoem Valdty of many esults n analyss depends on the completeness popety. Ths popety addesses the nadequacy of the system of atonal

More information

N = N t ; t 0. N is the number of claims paid by the

N = N t ; t 0. N is the number of claims paid by the Iulan MICEA, Ph Mhaela COVIG, Ph Canddate epatment of Mathematcs The Buchaest Academy of Economc Studes an CECHIN-CISTA Uncedt Tac Bank, Lugoj SOME APPOXIMATIONS USE IN THE ISK POCESS OF INSUANCE COMPANY

More information

Test 1 phy What mass of a material with density ρ is required to make a hollow spherical shell having inner radius r i and outer radius r o?

Test 1 phy What mass of a material with density ρ is required to make a hollow spherical shell having inner radius r i and outer radius r o? Test 1 phy 0 1. a) What s the pupose of measuement? b) Wte all fou condtons, whch must be satsfed by a scala poduct. (Use dffeent symbols to dstngush opeatons on ectos fom opeatons on numbes.) c) What

More information

4 Recursive Linear Predictor

4 Recursive Linear Predictor 4 Recusve Lnea Pedcto The man objectve of ths chapte s to desgn a lnea pedcto wthout havng a po knowledge about the coelaton popetes of the nput sgnal. In the conventonal lnea pedcto the known coelaton

More information

Approximate Abundance Histograms and Their Use for Genome Size Estimation

Approximate Abundance Histograms and Their Use for Genome Size Estimation J. Hlaváčová (Ed.): ITAT 2017 Poceedngs, pp. 27 34 CEUR Wokshop Poceedngs Vol. 1885, ISSN 1613-0073, c 2017 M. Lpovský, T. Vnař, B. Bejová Appoxmate Abundance Hstogams and The Use fo Genome Sze Estmaton

More information

Physics 2A Chapter 11 - Universal Gravitation Fall 2017

Physics 2A Chapter 11 - Universal Gravitation Fall 2017 Physcs A Chapte - Unvesal Gavtaton Fall 07 hese notes ae ve pages. A quck summay: he text boxes n the notes contan the esults that wll compse the toolbox o Chapte. hee ae thee sectons: the law o gavtaton,

More information

UNIT10 PLANE OF REGRESSION

UNIT10 PLANE OF REGRESSION UIT0 PLAE OF REGRESSIO Plane of Regesson Stuctue 0. Intoducton Ojectves 0. Yule s otaton 0. Plane of Regesson fo thee Vaales 0.4 Popetes of Resduals 0.5 Vaance of the Resduals 0.6 Summay 0.7 Solutons /

More information

Physics 11b Lecture #2. Electric Field Electric Flux Gauss s Law

Physics 11b Lecture #2. Electric Field Electric Flux Gauss s Law Physcs 11b Lectue # Electc Feld Electc Flux Gauss s Law What We Dd Last Tme Electc chage = How object esponds to electc foce Comes n postve and negatve flavos Conseved Electc foce Coulomb s Law F Same

More information

Distinct 8-QAM+ Perfect Arrays Fanxin Zeng 1, a, Zhenyu Zhang 2,1, b, Linjie Qian 1, c

Distinct 8-QAM+ Perfect Arrays Fanxin Zeng 1, a, Zhenyu Zhang 2,1, b, Linjie Qian 1, c nd Intenatonal Confeence on Electcal Compute Engneeng and Electoncs (ICECEE 15) Dstnct 8-QAM+ Pefect Aays Fanxn Zeng 1 a Zhenyu Zhang 1 b Lnje Qan 1 c 1 Chongqng Key Laboatoy of Emegency Communcaton Chongqng

More information

COMPLEMENTARY ENERGY METHOD FOR CURVED COMPOSITE BEAMS

COMPLEMENTARY ENERGY METHOD FOR CURVED COMPOSITE BEAMS ultscence - XXX. mcocd Intenatonal ultdscplnay Scentfc Confeence Unvesty of skolc Hungay - pl 06 ISBN 978-963-358-3- COPLEENTRY ENERGY ETHOD FOR CURVED COPOSITE BES Ákos József Lengyel István Ecsed ssstant

More information

Thermodynamics of solids 4. Statistical thermodynamics and the 3 rd law. Kwangheon Park Kyung Hee University Department of Nuclear Engineering

Thermodynamics of solids 4. Statistical thermodynamics and the 3 rd law. Kwangheon Park Kyung Hee University Department of Nuclear Engineering Themodynamcs of solds 4. Statstcal themodynamcs and the 3 d law Kwangheon Pak Kyung Hee Unvesty Depatment of Nuclea Engneeng 4.1. Intoducton to statstcal themodynamcs Classcal themodynamcs Statstcal themodynamcs

More information

Integral Vector Operations and Related Theorems Applications in Mechanics and E&M

Integral Vector Operations and Related Theorems Applications in Mechanics and E&M Dola Bagayoko (0) Integal Vecto Opeatons and elated Theoems Applcatons n Mechancs and E&M Ι Basc Defnton Please efe to you calculus evewed below. Ι, ΙΙ, andιιι notes and textbooks fo detals on the concepts

More information

Set of square-integrable function 2 L : function space F

Set of square-integrable function 2 L : function space F Set of squae-ntegable functon L : functon space F Motvaton: In ou pevous dscussons we have seen that fo fee patcles wave equatons (Helmholt o Schödnge) can be expessed n tems of egenvalue equatons. H E,

More information

Chapter 23: Electric Potential

Chapter 23: Electric Potential Chapte 23: Electc Potental Electc Potental Enegy It tuns out (won t show ths) that the tostatc foce, qq 1 2 F ˆ = k, s consevatve. 2 Recall, fo any consevatve foce, t s always possble to wte the wok done

More information

Event Shape Update. T. Doyle S. Hanlon I. Skillicorn. A. Everett A. Savin. Event Shapes, A. Everett, U. Wisconsin ZEUS Meeting, October 15,

Event Shape Update. T. Doyle S. Hanlon I. Skillicorn. A. Everett A. Savin. Event Shapes, A. Everett, U. Wisconsin ZEUS Meeting, October 15, Event Shape Update A. Eveett A. Savn T. Doyle S. Hanlon I. Skllcon Event Shapes, A. Eveett, U. Wsconsn ZEUS Meetng, Octobe 15, 2003-1 Outlne Pogess of Event Shapes n DIS Smla to publshed pape: Powe Coecton

More information

Learning the structure of Bayesian belief networks

Learning the structure of Bayesian belief networks Lectue 17 Leanng the stuctue of Bayesan belef netwoks Mlos Hauskecht mlos@cs.ptt.edu 5329 Sennott Squae Leanng of BBN Leanng. Leanng of paametes of condtonal pobabltes Leanng of the netwok stuctue Vaables:

More information

Amplifier Constant Gain and Noise

Amplifier Constant Gain and Noise Amplfe Constant Gan and ose by Manfed Thumm and Wene Wesbeck Foschungszentum Kalsuhe n de Helmholtz - Gemenschaft Unvestät Kalsuhe (TH) Reseach Unvesty founded 85 Ccles of Constant Gan (I) If s taken to

More information

Exact Simplification of Support Vector Solutions

Exact Simplification of Support Vector Solutions Jounal of Machne Leanng Reseach 2 (200) 293-297 Submtted 3/0; Publshed 2/0 Exact Smplfcaton of Suppot Vecto Solutons Tom Downs TD@ITEE.UQ.EDU.AU School of Infomaton Technology and Electcal Engneeng Unvesty

More information

LASER ABLATION ICP-MS: DATA REDUCTION

LASER ABLATION ICP-MS: DATA REDUCTION Lee, C-T A Lase Ablaton Data educton 2006 LASE ABLATON CP-MS: DATA EDUCTON Cn-Ty A. Lee 24 Septembe 2006 Analyss and calculaton of concentatons Lase ablaton analyses ae done n tme-esolved mode. A ~30 s

More information

Machine Learning. Spectral Clustering. Lecture 23, April 14, Reading: Eric Xing 1

Machine Learning. Spectral Clustering. Lecture 23, April 14, Reading: Eric Xing 1 Machne Leanng -7/5 7/5-78, 78, Spng 8 Spectal Clusteng Ec Xng Lectue 3, pl 4, 8 Readng: Ec Xng Data Clusteng wo dffeent ctea Compactness, e.g., k-means, mxtue models Connectvty, e.g., spectal clusteng

More information

Rigid Bodies: Equivalent Systems of Forces

Rigid Bodies: Equivalent Systems of Forces Engneeng Statcs, ENGR 2301 Chapte 3 Rgd Bodes: Equvalent Sstems of oces Intoducton Teatment of a bod as a sngle patcle s not alwas possble. In geneal, the se of the bod and the specfc ponts of applcaton

More information

Optimal System for Warm Standby Components in the Presence of Standby Switching Failures, Two Types of Failures and General Repair Time

Optimal System for Warm Standby Components in the Presence of Standby Switching Failures, Two Types of Failures and General Repair Time Intenatonal Jounal of ompute Applcatons (5 ) Volume 44 No, Apl Optmal System fo Wam Standby omponents n the esence of Standby Swtchng Falues, Two Types of Falues and Geneal Repa Tme Mohamed Salah EL-Shebeny

More information

Energy in Closed Systems

Energy in Closed Systems Enegy n Closed Systems Anamta Palt palt.anamta@gmal.com Abstact The wtng ndcates a beakdown of the classcal laws. We consde consevaton of enegy wth a many body system n elaton to the nvese squae law and

More information

Engineering Mechanics. Force resultants, Torques, Scalar Products, Equivalent Force systems

Engineering Mechanics. Force resultants, Torques, Scalar Products, Equivalent Force systems Engneeng echancs oce esultants, Toques, Scala oducts, Equvalent oce sstems Tata cgaw-hll Companes, 008 Resultant of Two oces foce: acton of one bod on anothe; chaacteed b ts pont of applcaton, magntude,

More information

(8) Gain Stage and Simple Output Stage

(8) Gain Stage and Simple Output Stage EEEB23 Electoncs Analyss & Desgn (8) Gan Stage and Smple Output Stage Leanng Outcome Able to: Analyze an example of a gan stage and output stage of a multstage amplfe. efeence: Neamen, Chapte 11 8.0) ntoducton

More information

Part V: Velocity and Acceleration Analysis of Mechanisms

Part V: Velocity and Acceleration Analysis of Mechanisms Pat V: Velocty an Acceleaton Analyss of Mechansms Ths secton wll evew the most common an cuently pactce methos fo completng the knematcs analyss of mechansms; escbng moton though velocty an acceleaton.

More information

A NOTE ON ELASTICITY ESTIMATION OF CENSORED DEMAND

A NOTE ON ELASTICITY ESTIMATION OF CENSORED DEMAND Octobe 003 B 003-09 A NOT ON ASTICITY STIATION OF CNSOD DAND Dansheng Dong an Hay. Kase Conell nvesty Depatment of Apple conomcs an anagement College of Agcultue an fe Scences Conell nvesty Ithaca New

More information

Vibration Input Identification using Dynamic Strain Measurement

Vibration Input Identification using Dynamic Strain Measurement Vbaton Input Identfcaton usng Dynamc Stan Measuement Takum ITOFUJI 1 ;TakuyaYOSHIMURA ; 1, Tokyo Metopoltan Unvesty, Japan ABSTRACT Tansfe Path Analyss (TPA) has been conducted n ode to mpove the nose

More information

V. Principles of Irreversible Thermodynamics. s = S - S 0 (7.3) s = = - g i, k. "Flux": = da i. "Force": = -Â g a ik k = X i. Â J i X i (7.

V. Principles of Irreversible Thermodynamics. s = S - S 0 (7.3) s = = - g i, k. Flux: = da i. Force: = -Â g a ik k = X i. Â J i X i (7. Themodynamcs and Knetcs of Solds 71 V. Pncples of Ievesble Themodynamcs 5. Onsage s Teatment s = S - S 0 = s( a 1, a 2,...) a n = A g - A n (7.6) Equlbum themodynamcs detemnes the paametes of an equlbum

More information

9/12/2013. Microelectronics Circuit Analysis and Design. Modes of Operation. Cross Section of Integrated Circuit npn Transistor

9/12/2013. Microelectronics Circuit Analysis and Design. Modes of Operation. Cross Section of Integrated Circuit npn Transistor Mcoelectoncs Ccut Analyss and Desgn Donald A. Neamen Chapte 5 The pola Juncton Tanssto In ths chapte, we wll: Dscuss the physcal stuctue and opeaton of the bpola juncton tanssto. Undestand the dc analyss

More information

Machine Learning 4771

Machine Learning 4771 Machne Leanng 4771 Instucto: Tony Jebaa Topc 6 Revew: Suppot Vecto Machnes Pmal & Dual Soluton Non-sepaable SVMs Kenels SVM Demo Revew: SVM Suppot vecto machnes ae (n the smplest case) lnea classfes that

More information

Chapter 8. Linear Momentum, Impulse, and Collisions

Chapter 8. Linear Momentum, Impulse, and Collisions Chapte 8 Lnea oentu, Ipulse, and Collsons 8. Lnea oentu and Ipulse The lnea oentu p of a patcle of ass ovng wth velocty v s defned as: p " v ote that p s a vecto that ponts n the sae decton as the velocty

More information

On a New Definition of a Stochastic-based Accuracy Concept of Data Reconciliation-Based Estimators

On a New Definition of a Stochastic-based Accuracy Concept of Data Reconciliation-Based Estimators On a New Defnton of a Stochastc-based Accuacy Concept of Data Reconclaton-Based Estmatos M. Bagajewcz Unesty of Olahoma 100 E. Boyd St., Noman OK 73019, USA Abstact Tadtonally, accuacy of an nstument s

More information

CSJM University Class: B.Sc.-II Sub:Physics Paper-II Title: Electromagnetics Unit-1: Electrostatics Lecture: 1 to 4

CSJM University Class: B.Sc.-II Sub:Physics Paper-II Title: Electromagnetics Unit-1: Electrostatics Lecture: 1 to 4 CSJM Unvesty Class: B.Sc.-II Sub:Physcs Pape-II Ttle: Electomagnetcs Unt-: Electostatcs Lectue: to 4 Electostatcs: It deals the study of behavo of statc o statonay Chages. Electc Chage: It s popety by

More information

On Maneuvering Target Tracking with Online Observed Colored Glint Noise Parameter Estimation

On Maneuvering Target Tracking with Online Observed Colored Glint Noise Parameter Estimation Wold Academy of Scence, Engneeng and Technology 6 7 On Maneuveng Taget Tacng wth Onlne Obseved Coloed Glnt Nose Paamete Estmaton M. A. Masnad-Sha, and S. A. Banan Abstact In ths pape a compehensve algothm

More information

CS649 Sensor Networks IP Track Lecture 3: Target/Source Localization in Sensor Networks

CS649 Sensor Networks IP Track Lecture 3: Target/Source Localization in Sensor Networks C649 enso etwoks IP Tack Lectue 3: Taget/ouce Localaton n enso etwoks I-Jeng Wang http://hng.cs.jhu.edu/wsn06/ png 006 C 649 Taget/ouce Localaton n Weless enso etwoks Basc Poblem tatement: Collaboatve

More information

19 The Born-Oppenheimer Approximation

19 The Born-Oppenheimer Approximation 9 The Bon-Oppenheme Appoxmaton The full nonelatvstc Hamltonan fo a molecule s gven by (n a.u.) Ĥ = A M A A A, Z A + A + >j j (883) Lets ewte the Hamltonan to emphasze the goal as Ĥ = + A A A, >j j M A

More information

24-2: Electric Potential Energy. 24-1: What is physics

24-2: Electric Potential Energy. 24-1: What is physics D. Iyad SAADEDDIN Chapte 4: Electc Potental Electc potental Enegy and Electc potental Calculatng the E-potental fom E-feld fo dffeent chage dstbutons Calculatng the E-feld fom E-potental Potental of a

More information

Economics 130. Lecture 4 Simple Linear Regression Continued

Economics 130. Lecture 4 Simple Linear Regression Continued Economcs 130 Lecture 4 Contnued Readngs for Week 4 Text, Chapter and 3. We contnue wth addressng our second ssue + add n how we evaluate these relatonshps: Where do we get data to do ths analyss? How do

More information

Generating Functions, Weighted and Non-Weighted Sums for Powers of Second-Order Recurrence Sequences

Generating Functions, Weighted and Non-Weighted Sums for Powers of Second-Order Recurrence Sequences Geneatng Functons, Weghted and Non-Weghted Sums fo Powes of Second-Ode Recuence Sequences Pantelmon Stăncă Aubun Unvesty Montgomey, Depatment of Mathematcs Montgomey, AL 3614-403, USA e-mal: stanca@studel.aum.edu

More information

Contact, information, consultations

Contact, information, consultations ontact, nfomaton, consultatons hemsty A Bldg; oom 07 phone: 058-347-769 cellula: 664 66 97 E-mal: wojtek_c@pg.gda.pl Offce hous: Fday, 9-0 a.m. A quote of the week (o camel of the week): hee s no expedence

More information

Remember: When an object falls due to gravity its potential energy decreases.

Remember: When an object falls due to gravity its potential energy decreases. Chapte 5: lectc Potental As mentoned seveal tmes dung the uate Newton s law o gavty and Coulomb s law ae dentcal n the mathematcal om. So, most thngs that ae tue o gavty ae also tue o electostatcs! Hee

More information

Dirichlet Mixture Priors: Inference and Adjustment

Dirichlet Mixture Priors: Inference and Adjustment Dchlet Mxtue Pos: Infeence and Adustment Xugang Ye (Wokng wth Stephen Altschul and Y Kuo Yu) Natonal Cante fo Botechnology Infomaton Motvaton Real-wold obects Independent obsevatons Categocal data () (2)

More information

A Study about One-Dimensional Steady State. Heat Transfer in Cylindrical and. Spherical Coordinates

A Study about One-Dimensional Steady State. Heat Transfer in Cylindrical and. Spherical Coordinates Appled Mathematcal Scences, Vol. 7, 03, no. 5, 67-633 HIKARI Ltd, www.m-hka.com http://dx.do.og/0.988/ams.03.38448 A Study about One-Dmensonal Steady State Heat ansfe n ylndcal and Sphecal oodnates Lesson

More information

Experimental study on parameter choices in norm-r support vector regression machines with noisy input

Experimental study on parameter choices in norm-r support vector regression machines with noisy input Soft Comput 006) 0: 9 3 DOI 0.007/s00500-005-0474-z ORIGINAL PAPER S. Wang J. Zhu F. L. Chung Hu Dewen Expemental study on paamete choces n nom- suppot vecto egesson machnes wth nosy nput Publshed onlne:

More information

Khintchine-Type Inequalities and Their Applications in Optimization

Khintchine-Type Inequalities and Their Applications in Optimization Khntchne-Type Inequaltes and The Applcatons n Optmzaton Anthony Man-Cho So Depatment of Systems Engneeng & Engneeng Management The Chnese Unvesty of Hong Kong ISDS-Kolloquum Unvestaet Wen 29 June 2009

More information

Physics Exam II Chapters 25-29

Physics Exam II Chapters 25-29 Physcs 114 1 Exam II Chaptes 5-9 Answe 8 of the followng 9 questons o poblems. Each one s weghted equally. Clealy mak on you blue book whch numbe you do not want gaded. If you ae not sue whch one you do

More information

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

Detection and Estimation Theory

Detection and Estimation Theory ESE 54 Detecton and Etmaton Theoy Joeph A. O Sullvan Samuel C. Sach Pofeo Electonc Sytem and Sgnal Reeach Laboatoy Electcal and Sytem Engneeng Wahngton Unvety 411 Jolley Hall 314-935-4173 (Lnda anwe) jao@wutl.edu

More information

ON THE FRESNEL SINE INTEGRAL AND THE CONVOLUTION

ON THE FRESNEL SINE INTEGRAL AND THE CONVOLUTION IJMMS 3:37, 37 333 PII. S16117131151 http://jmms.hndaw.com Hndaw Publshng Cop. ON THE FRESNEL SINE INTEGRAL AND THE CONVOLUTION ADEM KILIÇMAN Receved 19 Novembe and n evsed fom 7 Mach 3 The Fesnel sne

More information

Minimal Detectable Biases of GPS observations for a weighted ionosphere

Minimal Detectable Biases of GPS observations for a weighted ionosphere LETTER Eath Planets Space, 52, 857 862, 2000 Mnmal Detectable Bases of GPS obsevatons fo a weghted onosphee K. de Jong and P. J. G. Teunssen Depatment of Mathematcal Geodesy and Postonng, Delft Unvesty

More information

PHYS 705: Classical Mechanics. Derivation of Lagrange Equations from D Alembert s Principle

PHYS 705: Classical Mechanics. Derivation of Lagrange Equations from D Alembert s Principle 1 PHYS 705: Classcal Mechancs Devaton of Lagange Equatons fom D Alembet s Pncple 2 D Alembet s Pncple Followng a smla agument fo the vtual dsplacement to be consstent wth constants,.e, (no vtual wok fo

More information

Robust Feature Induction for Support Vector Machines

Robust Feature Induction for Support Vector Machines Robust Featue Inducton fo Suppot Vecto Machnes Rong Jn Depatment of Compute Scence and Engneeng, Mchgan State Unvesty, East Lansng, MI4884 ROGJI@CSE.MSU.EDU Huan Lu Depatment of Compute Scence and Engneeng,

More information

GENERALIZATION OF AN IDENTITY INVOLVING THE GENERALIZED FIBONACCI NUMBERS AND ITS APPLICATIONS

GENERALIZATION OF AN IDENTITY INVOLVING THE GENERALIZED FIBONACCI NUMBERS AND ITS APPLICATIONS #A39 INTEGERS 9 (009), 497-513 GENERALIZATION OF AN IDENTITY INVOLVING THE GENERALIZED FIBONACCI NUMBERS AND ITS APPLICATIONS Mohaad Faokh D. G. Depatent of Matheatcs, Fedows Unvesty of Mashhad, Mashhad,

More information

A. P. Sakis Meliopoulos Power System Modeling, Analysis and Control. Chapter 7 3 Operating State Estimation 3

A. P. Sakis Meliopoulos Power System Modeling, Analysis and Control. Chapter 7 3 Operating State Estimation 3 DRAF and INCOMPLEE able of Contents fom A. P. Saks Melopoulos Powe System Modelng, Analyss and Contol Chapte 7 3 Opeatng State Estmaton 3 7. Intoducton 3 7. SCADA System 4 7.3 System Netwok Confguato 7

More information

PARAMETER ESTIMATION FOR TWO WEIBULL POPULATIONS UNDER JOINT TYPE II CENSORED SCHEME

PARAMETER ESTIMATION FOR TWO WEIBULL POPULATIONS UNDER JOINT TYPE II CENSORED SCHEME Sept 04 Vol 5 No 04 Intenatonal Jounal of Engneeng Appled Scences 0-04 EAAS & ARF All ghts eseed wwweaas-ounalog ISSN305-869 PARAMETER ESTIMATION FOR TWO WEIBULL POPULATIONS UNDER JOINT TYPE II CENSORED

More information

Physics 202, Lecture 2. Announcements

Physics 202, Lecture 2. Announcements Physcs 202, Lectue 2 Today s Topcs Announcements Electc Felds Moe on the Electc Foce (Coulomb s Law The Electc Feld Moton of Chaged Patcles n an Electc Feld Announcements Homewok Assgnment #1: WebAssgn

More information

A Tutorial on Low Density Parity-Check Codes

A Tutorial on Low Density Parity-Check Codes A Tutoal on Low Densty Paty-Check Codes Tuan Ta The Unvesty of Texas at Austn Abstact Low densty paty-check codes ae one of the hottest topcs n codng theoy nowadays. Equpped wth vey fast encodng and decodng

More information

On the Distribution of the Product and Ratio of Independent Central and Doubly Non-central Generalized Gamma Ratio random variables

On the Distribution of the Product and Ratio of Independent Central and Doubly Non-central Generalized Gamma Ratio random variables On the Dstbuton of the Poduct Rato of Independent Cental Doubly Non-cental Genealzed Gamma Rato om vaables Calos A. Coelho João T. Mexa Abstact Usng a decomposton of the chaactestc functon of the logathm

More information

APPLICATIONS OF SEMIGENERALIZED -CLOSED SETS

APPLICATIONS OF SEMIGENERALIZED -CLOSED SETS Intenatonal Jounal of Mathematcal Engneeng Scence ISSN : 22776982 Volume Issue 4 (Apl 202) http://www.mes.com/ https://stes.google.com/ste/mesounal/ APPLICATIONS OF SEMIGENERALIZED CLOSED SETS G.SHANMUGAM,

More information

4 SingularValue Decomposition (SVD)

4 SingularValue Decomposition (SVD) /6/00 Z:\ jeh\self\boo Kannan\Jan-5-00\4 SVD 4 SngulaValue Decomposton (SVD) Chapte 4 Pat SVD he sngula value decomposton of a matx s the factozaton of nto the poduct of thee matces = UDV whee the columns

More information

E For K > 0. s s s s Fall Physical Chemistry (II) by M. Lim. singlet. triplet

E For K > 0. s s s s Fall Physical Chemistry (II) by M. Lim. singlet. triplet Eneges of He electonc ψ E Fo K > 0 ψ = snglet ( )( ) s s+ ss αβ E βα snglet = ε + ε + J s + Ks Etplet = ε + ε + J s Ks αα ψ tplet = ( s s ss ) ββ ( αβ + βα ) s s s s s s s s ψ G = ss( αβ βα ) E = ε + ε

More information

VParC: A Compression Scheme for Numeric Data in Column-Oriented Databases

VParC: A Compression Scheme for Numeric Data in Column-Oriented Databases The Intenatonal Aab Jounal of Infomaton Technology VPaC: A Compesson Scheme fo Numec Data n Column-Oented Databases Ke Yan, Hong Zhu, and Kevn Lü School of Compute Scence and Technology, Huazhong Unvesty

More information

Professor Wei Zhu. 1. Sampling from the Normal Population

Professor Wei Zhu. 1. Sampling from the Normal Population AMS570 Pofesso We Zhu. Samplg fom the Nomal Populato *Example: We wsh to estmate the dstbuto of heghts of adult US male. It s beleved that the heght of adult US male follows a omal dstbuto N(, ) Def. Smple

More information

Pattern Analyses (EOF Analysis) Introduction Definition of EOFs Estimation of EOFs Inference Rotated EOFs

Pattern Analyses (EOF Analysis) Introduction Definition of EOFs Estimation of EOFs Inference Rotated EOFs Patten Analyses (EOF Analyss) Intoducton Defnton of EOFs Estmaton of EOFs Infeence Rotated EOFs . Patten Analyses Intoducton: What s t about? Patten analyses ae technques used to dentfy pattens of the

More information

Lab 10: Newton s Second Law in Rotation

Lab 10: Newton s Second Law in Rotation Lab 10: Newton s Second Law in Rotation We can descibe the motion of objects that otate (i.e. spin on an axis, like a popelle o a doo) using the same definitions, adapted fo otational motion, that we have

More information

CHAPTER 7. Multivariate effect sizes indices

CHAPTER 7. Multivariate effect sizes indices CHAPTE 7 Multvaate effect szes ndces Seldom does one fnd that thee s only a sngle dependent vaable nvolved n a study. In Chapte 3 s Example A we have the vaables BDI, POMS_S and POMS_B, n Example E thee

More information

ASTR415: Problem Set #6

ASTR415: Problem Set #6 ASTR45: Poblem Set #6 Cuan D. Muhlbege Univesity of Mayland (Dated: May 7, 27) Using existing implementations of the leapfog and Runge-Kutta methods fo solving coupled odinay diffeential equations, seveal

More information

Correspondence Analysis & Related Methods

Correspondence Analysis & Related Methods Coespondence Analyss & Related Methods Ineta contbutons n weghted PCA PCA s a method of data vsualzaton whch epesents the tue postons of ponts n a map whch comes closest to all the ponts, closest n sense

More information

State Estimation. Ali Abur Northeastern University, USA. Nov. 01, 2017 Fall 2017 CURENT Course Lecture Notes

State Estimation. Ali Abur Northeastern University, USA. Nov. 01, 2017 Fall 2017 CURENT Course Lecture Notes State Estmaton Al Abu Notheasten Unvesty, USA Nov. 0, 07 Fall 07 CURENT Couse Lectue Notes Opeatng States of a Powe System Al Abu NORMAL STATE SECURE o INSECURE RESTORATIVE STATE EMERGENCY STATE PARTIAL

More information

Some Approximate Analytical Steady-State Solutions for Cylindrical Fin

Some Approximate Analytical Steady-State Solutions for Cylindrical Fin Some Appoxmate Analytcal Steady-State Solutons fo Cylndcal Fn ANITA BRUVERE ANDRIS BUIIS Insttute of Mathematcs and Compute Scence Unvesty of Latva Rana ulv 9 Rga LV459 LATVIA Astact: - In ths pape we

More information

Slide 1. Quantum Mechanics: the Practice

Slide 1. Quantum Mechanics: the Practice Slde Quantum Mecancs: te Pactce Slde Remnde: Electons As Waves Wavelengt momentum = Planck? λ p = = 6.6 x 0-34 J s Te wave s an exctaton a vbaton: We need to know te ampltude of te exctaton at evey pont

More information

A Novel Ordinal Regression Method with Minimum Class Variance Support Vector Machine

A Novel Ordinal Regression Method with Minimum Class Variance Support Vector Machine Intenatonal Confeence on Mateals Engneeng and Infomaton echnology Applcatons (MEIA 05) A ovel Odnal Regesson Method wth Mnmum Class Vaance Suppot Vecto Machne Jnong Hu,, a, Xaomng Wang and Zengx Huang

More information

Chapter 2 - The Simple Linear Regression Model S =0. e i is a random error. S β2 β. This is a minimization problem. Solution is a calculus exercise.

Chapter 2 - The Simple Linear Regression Model S =0. e i is a random error. S β2 β. This is a minimization problem. Solution is a calculus exercise. Chapter - The Smple Lnear Regresson Model The lnear regresson equaton s: where y + = β + β e for =,..., y and are observable varables e s a random error How can an estmaton rule be constructed for the

More information

CEEP-BIT WORKING PAPER SERIES. Efficiency evaluation of multistage supply chain with data envelopment analysis models

CEEP-BIT WORKING PAPER SERIES. Efficiency evaluation of multistage supply chain with data envelopment analysis models CEEP-BIT WORKING PPER SERIES Effcency evaluaton of multstage supply chan wth data envelopment analyss models Ke Wang Wokng Pape 48 http://ceep.bt.edu.cn/englsh/publcatons/wp/ndex.htm Cente fo Enegy and

More information

e i is a random error

e i is a random error Chapter - The Smple Lnear Regresson Model The lnear regresson equaton s: where + β + β e for,..., and are observable varables e s a random error How can an estmaton rule be constructed for the unknown

More information

A-Kappa: A measure of Agreement among Multiple Raters

A-Kappa: A measure of Agreement among Multiple Raters Jounal of Data Scence (04), 697-76 A-Kappa: A measue of Ageement among Multple Rates Shva Gautam Beth Isael Deaconess Medcal Cente, Havad Medcal School Abstact: Medcal data and bomedcal studes ae often

More information

Bayesian Assessment of Availabilities and Unavailabilities of Multistate Monotone Systems

Bayesian Assessment of Availabilities and Unavailabilities of Multistate Monotone Systems Dept. of Math. Unvesty of Oslo Statstcal Reseach Repot No 3 ISSN 0806 3842 June 2010 Bayesan Assessment of Avalabltes and Unavalabltes of Multstate Monotone Systems Bent Natvg Jøund Gåsemy Tond Retan June

More information

2/24/2014. The point mass. Impulse for a single collision The impulse of a force is a vector. The Center of Mass. System of particles

2/24/2014. The point mass. Impulse for a single collision The impulse of a force is a vector. The Center of Mass. System of particles /4/04 Chapte 7 Lnea oentu Lnea oentu of a Sngle Patcle Lnea oentu: p υ It s a easue of the patcle s oton It s a vecto, sla to the veloct p υ p υ p υ z z p It also depends on the ass of the object, sla

More information

Evaluation of Various Types of Wall Boundary Conditions for the Boltzmann Equation

Evaluation of Various Types of Wall Boundary Conditions for the Boltzmann Equation Ealuaton o Vaous Types o Wall Bounday Condtons o the Boltzmann Equaton Chstophe D. Wlson a, Ramesh K. Agawal a, and Felx G. Tcheemssne b a Depatment o Mechancal Engneeng and Mateals Scence Washngton Unesty

More information

Lecture 4 Hypothesis Testing

Lecture 4 Hypothesis Testing Lecture 4 Hypothess Testng We may wsh to test pror hypotheses about the coeffcents we estmate. We can use the estmates to test whether the data rejects our hypothess. An example mght be that we wsh to

More information

Mechanics Physics 151

Mechanics Physics 151 Mechancs Physcs 151 Lectue 18 Hamltonan Equatons of Moton (Chapte 8) What s Ahead We ae statng Hamltonan fomalsm Hamltonan equaton Today and 11/6 Canoncal tansfomaton 1/3, 1/5, 1/10 Close lnk to non-elatvstc

More information

Basically, if you have a dummy dependent variable you will be estimating a probability.

Basically, if you have a dummy dependent variable you will be estimating a probability. ECON 497: Lecture Notes 13 Page 1 of 1 Metropoltan State Unversty ECON 497: Research and Forecastng Lecture Notes 13 Dummy Dependent Varable Technques Studenmund Chapter 13 Bascally, f you have a dummy

More information

33. 12, or its reciprocal. or its negative.

33. 12, or its reciprocal. or its negative. Page 6 The Point is Measuement In spite of most of what has been said up to this point, we did not undetake this poject with the intent of building bette themometes. The point is to measue the peson. Because

More information

Using DP for hierarchical discretization of continuous attributes. Amit Goyal (31 st March 2008)

Using DP for hierarchical discretization of continuous attributes. Amit Goyal (31 st March 2008) Usng DP fo heachcal dscetzaton of contnos attbtes Amt Goyal 31 st Mach 2008 Refeence Chng-Cheng Shen and Yen-Lang Chen. A dynamc-pogammng algothm fo heachcal dscetzaton of contnos attbtes. In Eopean Jonal

More information

iclicker Quiz a) True b) False Theoretical physics: the eternal quest for a missing minus sign and/or a factor of two. Which will be an issue today?

iclicker Quiz a) True b) False Theoretical physics: the eternal quest for a missing minus sign and/or a factor of two. Which will be an issue today? Clce Quz I egsteed my quz tansmtte va the couse webste (not on the clce.com webste. I ealze that untl I do so, my quz scoes wll not be ecoded. a Tue b False Theoetcal hyscs: the etenal quest fo a mssng

More information

A Queuing Model for an Automated Workstation Receiving Jobs from an Automated Workstation

A Queuing Model for an Automated Workstation Receiving Jobs from an Automated Workstation Intenatonal Jounal of Opeatons Reseach Intenatonal Jounal of Opeatons Reseach Vol. 7, o. 4, 918 (1 A Queung Model fo an Automated Wokstaton Recevng Jobs fom an Automated Wokstaton Davd S. Km School of

More information

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

More information